Patient Matching and HIE: Do We Even Have a Strategy?
Accurate patient matching continues to be a challenge in healthcare, especially in an HIE environment (see my earlier HIMSS posting). No simple solution appears to be at hand. The issue is also inextricably linked to issues of patient identity which are also difficult to solve at our current level of technology. Several approaches have emerged, however, which do not yet seem to be converging. I would characterize the major approaches as follows:
Traditional Approach: In the absence of anything better or more practical, the traditional approach is to use matching algorithms – both deterministic and probabilistic – to examine patient data at a granular level and hope that both false positive and false negative matches can be avoided. Various organizations have different tolerance levels for error, some advocating little or no tolerance at all (See ONC Nationwide Interoperability Roadmap, p. 36). Public health agencies have been working successfully with these algorithms for many years, and there are a number of Open Source and commercial products in this space.
Biometrics: For proponents of this approach, the best way to identify someone is by including a personal attribute in a patient’s data that can’t be faked – a biometric of one sort or another – to identify a patient consistently across care settings. Anything short of that opens up the process to unacceptable error. But biometrics can be faked and in many settings today the hardware and software is not in place to support their capture, safe storage, and exchange.
Unique identifier: These folks believe that only through the adoption of single, national patient identifier will we be able to consistently and longitudinally match our distributed patient records across provider sites and settings. They do not seem deterred by the current Congressional ban on the Federal Government from promulgating or adopting such a strategy (see Public Law 105–277. 105th Congress. October 21, 1998). In some cases proponents of this strategy are advocating for a voluntary unique identifier assigned by a neutral organization (for an example see https://www.gpii.info/).
Health Record Bank (HRB): For these folks, patient matching should be solved by, well, the patient (see http://healthbanking.org/). Medical records should be aggregated by the patient in a central repository, or bank, much as financial assets are accumulated there and made available by the patient for authorized transactions or uses. Though this is an intriguing notion, HRBs have failed to generate any traction in the marketplace beyond some very limited experiments.
Active work is going on in all of these strategies, with no clear cut winner as of yet. The public and private sector need to get together to discuss and pilot various approaches and to encourage Congress to reexamine its current position.